Search

This paper gives an information hiding method for H.264/AVC compressed video stream based on integer transform coefficient and virtual space encoding. In this method, the integer transform coefficients which are selected as carriers are divided into multiple groups. And it only needs to change at most the value of one of the coefficients in each group to embed $$\hbox {Log}_{2}(2N+1)$$ binary bits when each group contains $$N$$ coefficients. Moreover, the required change is simple, which can be achieved by adding 1 or $$-$$ 1 to the coefficient. Since only one coefficient's value is slightly changed for every $$\hbox {Log}_{2}(2N+1)$$ secret bits, this method does less harm to the original encoded video stream caused by the information embedding compared with that by current methods. Furthermore, this method only chooses the macroblock that has rich texture as the candidate carrier macroblock, and it employs the embedding position template to select the candidate $$4\times 4$$ coded blocks (I4-blocks) from the carrier macroblock for watermark bits embedding. In this way, the counteracting steganalysis performance is significantly improved. Experimental results on several test sequences demonstrate that the proposed approach can realize blind extraction and the performance is superior to that of the other current state-of-the-art information hiding methods also based on modulating integer transform coefficients. [ABSTRACT FROM AUTHOR]